It’s been five years since I collected with industry thought leaders in Reykjavik, in the cold and the dark of an Icelandic winter. In that first 2020 Executive Summit, we together laid-out the vision for a digitized and automated network, and by the title of our meeting, gave ourselves five years to see the vision through.
Back then, everyone was buzzing about the transformation potential of network digitization. The status quo was about to get upended. Communications service providers (CSPs) would finally break free from the spiral of escalating costs and shrinking margins, battling for the same static pool of subscribers. We would digitize and automate our way out. The 5G applications on the horizon—massive IoT connectivity, low-latency gaming, autonomous vehicles and others—would spawn explosive growth in network connections and new business models and revenue streams.
As I prepare to collect with the group again, the clock is up, so it’s surely fair to ask: How much progress have we made since 2014? How close are we to actually realizing that transformation? Unfortunately, in many cases, the answers to these questions are “very little” and “not so much”.
To be fair, the deep technology delivery of 5G—new radios, new handsets—have proceeded broadly on schedule. But at the core of these networks, progress remains moribund. This is very troubling because this new network was never supposed to be just about faster handsets. Progress at the core is critical if we are to adopt the multi-billions of new devices that bring online everything from wind farms to refrigerators to cars.
What’s taking so long? If we all know where CSP networks need to go, what’s stopping us from getting there? And what can we do to fix it?
A Massive 5G Technology Deficit
Delivering faster data rates is well and good, but the real industry transformation—the real money for CSPs—derives from the dynamic real-time capabilities of 5G networks. The ability to position resources anywhere they’re needed at the edge. To deliver networks finely tuned to specific use cases and industries, that adapt in real time to the needs of applications running on them.
We’ve nailed the data rates. But accomplishing the rest requires a transition from fixed-configuration to dynamic-configuration systems. And CSPs cannot get there from where they are today, because they simply haven’t laid the groundwork for automating their networks.
Right now, just provisioning a new circuit between cities can take months, even when all the equipment is in place. Requests pass from one domain-specific operations team to the next, each existing on its own island. CSPs can’t even visualize their networks from the perspective of end-to-end services, much less control them that way.
As we contemplate a 10x increase in the number of radio locations and equipment with 5G, like a juggler managing ever-more spinning plates, we move from the difficult to the impossible. And now, CSPs are supposed to launch new services that depend on an intent-driven network recognizing what’s needed and reacting in real time? Good luck.
Barriers to Dynamic Automation
The need for unified management isn’t a mystery—CSPs have been talking about it for years. So why don’t they have it?
- Antiquated vendor models—Today, the telco vendor market looks a lot like the mini-computer business of the mid-90s. There’s a handful of big vendors pushing proprietary systems, with CSPs hoping that if they just buy everything from one vendor, it will all work out. But no single vendor covers all parts of the network involved in an end-to-end service. So you end up with islands of domain-specific equipment and management systems, none of which can talk to each other or be unified within a single framework.
- False starts and wasted cycles—Along these lines, network device vendors have pushed what amounts to a pipe dream that they can solve these problems using overlay networks and handling dynamic service creation at the application layer. But there is no magical overlay that lets one vendor control another’s equipment when everyone is jealously guarding their own turf. Worse, the overlay becomes just one more thing to manage on top of all the other domains!
- Shortsighted carrier focus—CSPs themselves still think about network management in terms of domains rather than services. For years, carriers didn’t even pay for management tools, they just used what vendors gave them. Now that they see how those proprietary tools lock them in, they’re realizing how much work it takes to bring everything into a unified framework. Who funds that? It’s not a line item they’ve had to budget for before.
- Lack of open standards and interfaces—Going back to the minicomputer example, what finally launched the modern era of computing was the rise of Linux—an open, standardized computing approach, where you could be confident your software would work across any vendor’s hardware. What many have forgotten is that what enabled Linux was POSIX–standards that enabled applications to be ported between UNIX distributions. In networking, we’ve seen excellent work from open source communities in the telco space—projects like OpenDaylight, OpenStack and ONAP, supported by standards such as NETCONF and YANG. But CSPs have had to drag vendors kicking and screaming to support them, and not all vendors seem to have gotten the memo.
- Lack of startup innovation—As I wrote in a recent blog, when industries hit a wall like this in the past, new startups were ready to solve the problems that entrenched vendors couldn’t. Where is that startup innovation today? Record VC funding is going to SD-WAN, autonomous vehicles, and dozens of other applications that will ride on CSP networks. Almost none of it’s going toward telco infrastructure itself because of the notoriously long sales cycles and conservative mindset of CSPs.
It’s Time to Start Paying that Technical Debt
Clearly, CSPs still have big hurdles to overcome before we see the real industry transformation everyone’s banking on. Some CSPs are already using open systems to automate provisioning in production networks, but they remain a small minority. Most are sitting on their hands, waiting for vendors to make 5G experiences possible for them. Effectively, they’re trying to outsource the transformation of their business, and that never works.
If we’re going to see real change, CSPs have to make it happen themselves. Here’s how they can get started:
- Start automating—Before you can consider new 5G experiences, start by unifying existing multi-domain services. Start anywhere—backhaul, point-to-point connectivity, E-LINE. Bring all the different pieces together that service touches, so you can provision in minutes instead of weeks. Then, as you start thinking about things like edge compute, you can actually link it to your RAN, central offices, everywhere else in your network in a more automated way.
- Use open source—Open source models like OpenDaylight are the only way to break vendor lock-in and unify siloed network domains and devices. It’s the only common ground where vendors and CSPs can meet. But until more CSPs start actually using these tools in their networks, they won’t gain the critical mass needed to solve the problems we need to solve.
- Invest in startups—The innovation we need will only come when the world’s most promising young companies are attacking CSP network problems in creative new ways. But startups and investors won’t waste their time on telco infrastructure if telcos themselves aren’t committed to funding it. Just dipping your toes in isn’t enough. If you spend a dollar on startups and a thousand on incumbents, that’s not a real shift in your strategy.
We can get to real industry transformation, and 5G can be the catalyst to make it happen. But we have some catching up to do before we get there. It’s time for CSPs to take ownership of the technical deficit at the heart of their networks and take concrete steps to address it. Let’s not meet again in five years to lament on missing the transformation opportunity for 6G!
Here are some other informative resources for 5G and your digital transformation:
Downloadable eBook: Digital Transformation for 5G in the Real World
Infographic: The 5 Factors of 5G infographic
Video: Open Source in 5G – Lessons from Deployment